/* Copyright 2014 CoreOS, Inc. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. */ package raft import ( "errors" "log" "reflect" "github.com/coreos/etcd/Godeps/_workspace/src/code.google.com/p/go.net/context" pb "github.com/coreos/etcd/raft/raftpb" ) var ( emptyState = pb.HardState{} // ErrStopped is returned by methods on Nodes that have been stopped. ErrStopped = errors.New("raft: stopped") ) // SoftState provides state that is useful for logging and debugging. // The state is volatile and does not need to be persisted to the WAL. type SoftState struct { Lead uint64 RaftState StateType Nodes []uint64 } func (a *SoftState) equal(b *SoftState) bool { return reflect.DeepEqual(a, b) } // Ready encapsulates the entries and messages that are ready to read, // be saved to stable storage, committed or sent to other peers. // All fields in Ready are read-only. type Ready struct { // The current volatile state of a Node. // SoftState will be nil if there is no update. // It is not required to consume or store SoftState. *SoftState // The current state of a Node to be saved to stable storage BEFORE // Messages are sent. // HardState will be equal to empty state if there is no update. pb.HardState // Entries specifies entries to be saved to stable storage BEFORE // Messages are sent. Entries []pb.Entry // Snapshot specifies the snapshot to be saved to stable storage. Snapshot pb.Snapshot // CommittedEntries specifies entries to be committed to a // store/state-machine. These have previously been committed to stable // store. CommittedEntries []pb.Entry // Messages specifies outbound messages to be sent AFTER Entries are // committed to stable storage. Messages []pb.Message } type compact struct { index uint64 nodes []uint64 data []byte } func isHardStateEqual(a, b pb.HardState) bool { return a.Term == b.Term && a.Vote == b.Vote && a.Commit == b.Commit } // IsEmptyHardState returns true if the given HardState is empty. func IsEmptyHardState(st pb.HardState) bool { return isHardStateEqual(st, emptyState) } // IsEmptySnap returns true if the given Snapshot is empty. func IsEmptySnap(sp pb.Snapshot) bool { return sp.Index == 0 } func (rd Ready) containsUpdates() bool { return rd.SoftState != nil || !IsEmptyHardState(rd.HardState) || !IsEmptySnap(rd.Snapshot) || len(rd.Entries) > 0 || len(rd.CommittedEntries) > 0 || len(rd.Messages) > 0 } // Node represents a node in a raft cluster. type Node interface { // Tick increments the internal logical clock for the Node by a single tick. Election // timeouts and heartbeat timeouts are in units of ticks. Tick() // Campaign causes the Node to transition to candidate state and start campaigning to become leader. Campaign(ctx context.Context) error // Propose proposes that data be appended to the log. Propose(ctx context.Context, data []byte) error // ProposeConfChange proposes config change. // At most one ConfChange can be in the process of going through consensus. // Application needs to call ApplyConfChange when applying EntryConfChange type entry. ProposeConfChange(ctx context.Context, cc pb.ConfChange) error // Step advances the state machine using the given message. ctx.Err() will be returned, if any. Step(ctx context.Context, msg pb.Message) error // Ready returns a channel that returns the current point-in-time state // Users of the Node must call Advance after applying the state returned by Ready Ready() <-chan Ready // Advance notifies the Node that the application has applied and saved progress up to the last Ready. // It prepares the node to return the next available Ready. Advance() // ApplyConfChange applies config change to the local node. // TODO: reject existing node when add node // TODO: reject non-existant node when remove node ApplyConfChange(cc pb.ConfChange) // Stop performs any necessary termination of the Node Stop() // Compact discards the entrire log up to the given index. It also // generates a raft snapshot containing the given nodes configuration // and the given snapshot data. // It is the caller's responsibility to ensure the given configuration // and snapshot data match the actual point-in-time configuration and snapshot // at the given index. Compact(index uint64, nodes []uint64, d []byte) } type Peer struct { ID uint64 Context []byte } // StartNode returns a new Node given a unique raft id, a list of raft peers, and // the election and heartbeat timeouts in units of ticks. // It also builds ConfChangeAddNode entry for each peer and puts them at the head of the log. func StartNode(id uint64, peers []Peer, election, heartbeat int) Node { n := newNode() r := newRaft(id, nil, election, heartbeat) for _, peer := range peers { cc := pb.ConfChange{Type: pb.ConfChangeAddNode, NodeID: peer.ID, Context: peer.Context} d, err := cc.Marshal() if err != nil { panic("unexpected marshal error") } e := pb.Entry{Type: pb.EntryConfChange, Term: 1, Index: r.raftLog.lastIndex() + 1, Data: d} r.raftLog.append(r.raftLog.lastIndex(), e) } r.raftLog.committed = r.raftLog.lastIndex() go n.run(r) return &n } // RestartNode is identical to StartNode but takes an initial State and a slice // of entries. Generally this is used when restarting from a stable storage // log. func RestartNode(id uint64, election, heartbeat int, snapshot *pb.Snapshot, st pb.HardState, ents []pb.Entry) Node { n := newNode() r := newRaft(id, nil, election, heartbeat) if snapshot != nil { r.restore(*snapshot) } if !isHardStateEqual(st, emptyState) { r.loadState(st) } if len(ents) != 0 { r.loadEnts(ents) } go n.run(r) return &n } // node is the canonical implementation of the Node interface type node struct { propc chan pb.Message recvc chan pb.Message compactc chan compact confc chan pb.ConfChange readyc chan Ready advancec chan struct{} tickc chan struct{} done chan struct{} } func newNode() node { return node{ propc: make(chan pb.Message), recvc: make(chan pb.Message), compactc: make(chan compact), confc: make(chan pb.ConfChange), readyc: make(chan Ready), advancec: make(chan struct{}), tickc: make(chan struct{}), done: make(chan struct{}), } } func (n *node) Stop() { close(n.done) } func (n *node) run(r *raft) { var propc chan pb.Message var readyc chan Ready var advancec chan struct{} var prevLastUnstablei uint64 var rd Ready lead := None prevSoftSt := r.softState() prevHardSt := r.HardState prevSnapi := r.raftLog.snapshot.Index for { if advancec != nil { readyc = nil } else { rd = newReady(r, prevSoftSt, prevHardSt, prevSnapi) if rd.containsUpdates() { readyc = n.readyc } else { readyc = nil } if rd.SoftState != nil && lead != rd.SoftState.Lead { if r.hasLeader() { if lead == None { log.Printf("raft: elected leader %x at term %d", rd.SoftState.Lead, r.Term) } else { log.Printf("raft: leader changed from %x to %x at term %d", lead, rd.SoftState.Lead, r.Term) } propc = n.propc } else { log.Printf("raft: lost leader %x at term %d", lead, r.Term) propc = nil } lead = rd.SoftState.Lead } } select { // TODO: maybe buffer the config propose if there exists one (the way // described in raft dissertation) // Currently it is dropped in Step silently. case m := <-propc: m.From = r.id r.Step(m) case m := <-n.recvc: r.Step(m) // raft never returns an error case c := <-n.compactc: r.compact(c.index, c.nodes, c.data) case cc := <-n.confc: if cc.NodeID == None { r.resetPendingConf() break } switch cc.Type { case pb.ConfChangeAddNode: r.addNode(cc.NodeID) case pb.ConfChangeRemoveNode: r.removeNode(cc.NodeID) default: panic("unexpected conf type") } case <-n.tickc: r.tick() case readyc <- rd: if rd.SoftState != nil { prevSoftSt = rd.SoftState } if len(rd.Entries) > 0 { prevLastUnstablei = rd.Entries[len(rd.Entries)-1].Index } if !IsEmptyHardState(rd.HardState) { prevHardSt = rd.HardState } if !IsEmptySnap(rd.Snapshot) { prevSnapi = rd.Snapshot.Index if prevSnapi > prevLastUnstablei { prevLastUnstablei = prevSnapi } } r.msgs = nil advancec = n.advancec case <-advancec: if prevHardSt.Commit != 0 { r.raftLog.appliedTo(prevHardSt.Commit) } if prevLastUnstablei != 0 { r.raftLog.stableTo(prevLastUnstablei) } advancec = nil case <-n.done: return } } } // Tick increments the internal logical clock for this Node. Election timeouts // and heartbeat timeouts are in units of ticks. func (n *node) Tick() { select { case n.tickc <- struct{}{}: case <-n.done: } } func (n *node) Campaign(ctx context.Context) error { return n.step(ctx, pb.Message{Type: pb.MsgHup}) } func (n *node) Propose(ctx context.Context, data []byte) error { return n.step(ctx, pb.Message{Type: pb.MsgProp, Entries: []pb.Entry{{Data: data}}}) } func (n *node) Step(ctx context.Context, m pb.Message) error { // ignore unexpected local messages receiving over network if m.Type == pb.MsgHup || m.Type == pb.MsgBeat { // TODO: return an error? return nil } return n.step(ctx, m) } func (n *node) ProposeConfChange(ctx context.Context, cc pb.ConfChange) error { data, err := cc.Marshal() if err != nil { return err } return n.Step(ctx, pb.Message{Type: pb.MsgProp, Entries: []pb.Entry{{Type: pb.EntryConfChange, Data: data}}}) } // Step advances the state machine using msgs. The ctx.Err() will be returned, // if any. func (n *node) step(ctx context.Context, m pb.Message) error { ch := n.recvc if m.Type == pb.MsgProp { ch = n.propc } select { case ch <- m: return nil case <-ctx.Done(): return ctx.Err() case <-n.done: return ErrStopped } } func (n *node) Ready() <-chan Ready { return n.readyc } func (n *node) Advance() { select { case n.advancec <- struct{}{}: case <-n.done: } } func (n *node) ApplyConfChange(cc pb.ConfChange) { select { case n.confc <- cc: case <-n.done: } } func (n *node) Compact(index uint64, nodes []uint64, d []byte) { select { case n.compactc <- compact{index, nodes, d}: case <-n.done: } } func newReady(r *raft, prevSoftSt *SoftState, prevHardSt pb.HardState, prevSnapi uint64) Ready { rd := Ready{ Entries: r.raftLog.unstableEnts(), CommittedEntries: r.raftLog.nextEnts(), Messages: r.msgs, } if softSt := r.softState(); !softSt.equal(prevSoftSt) { rd.SoftState = softSt } if !isHardStateEqual(r.HardState, prevHardSt) { rd.HardState = r.HardState } if prevSnapi != r.raftLog.snapshot.Index { rd.Snapshot = r.raftLog.snapshot } return rd }